Blog Post

Copilot Studio Blog
3 MIN READ

More powerful agents and workflows for autonomous business processes: Introducing a new harness for Copilot Studio

Ryan_Cunningham's avatar
Aug 03, 2026

Editors' note: The GitHub Copilot harness is now the Microsoft Copilot harness. Learn more.

Over the past two months, we’ve been previewing a new capability within Copilot Studio that lets you build agents capable of taking on more complex business processes. Today, we’re excited to share that this capability is now generally available for production use and to introduce its new name: the GitHub Copilot harness in Copilot Studio. 

 

Meet the GitHub Copilot harness  

The GitHub Copilot harness gives Copilot Studio the coding and reasoning capabilities behind our most advanced agent experiences (like Copilot Cowork and the GitHub Copilot coding agent). This means your agents can now handle work that used to be out of reach, including processes that have many steps, many sources, and ambiguous decision points. 

The new harness is built for complex, long-horizon work using the latest frontier reasoning models like Opus 5, GPT-5.6 Sol, and Fable 5. It can plan, reason through dynamic problems, run an agentic loop, use skills, integrate workflows, connect to tools and agents in other platforms, and produce rich, multi-part outputs.  

In our own testing with real world business process evals, Copilot Studio shows significant performance and quality gains when using the GitHub Copilot harness. Improvements in multi-tool use, file analysis, code analysis, and knowledge quality mean these agents are more capable of taking on your most complex business processes. 

 

To help you build these more capable agents, we’re also delivering new experiences for makers building agents with the GitHub Copilot harness. The agent designer is more intuitive for authoring, putting the most important tools right within reach to make authoring faster, while retaining full agent lifecycle management features.  

A gif showing the process of adding a skill to an agent in the agent designer and starting a preview test conversation

Then, the workflow designer gives a visual canvas to understand and edit workflows, including adding agent nodes and running workflow evals. And soon, natural language authoring will let you describe your business goal and assemble the right combination of agents and workflows through a multi-turn conversation. 

Agents running on the GitHub Copilot harness use usage-based billing for all work, regardless of Microsoft 365 Copilot licensing. You pay for your agent’s usage based on the models you choose, the organizational context and tools you add, and runtime used. Certain AI-driven maker experiences, like natural language authoring, evaluations, and testing, will also fall under usage-based billing if building with the GitHub Copilot harness.  

You can learn more about Copilot Studio usage-based billing here. 

Continued Support for Copilot Chat and Standard harnesses 

Adding a new harness does not change the ones you already rely on. Different harnesses can be optimized for different outcomes, so rather than relying on a single one-size-fits-all approach, Copilot Studio now lets you choose the harness that best fits your scenario.  

This means Copilot Studio supports three harnesses today: 

  1. The Copilot Chat harness, which uses the same harness as Microsoft 365 Copilot Chat and is ideal for customizing Copilot Chat experiences.

  2. The Standard harness, which most agents built in Copilot Studio use today and is great for conversational agents with rules-based topics.

  3. The new GitHub Copilot harness, which uses the power of GitHub Copilot SDK to automate complex, agentic business processes. 

We will continue to support the Copilot Chat and Standard harnesses in Copilot Studio for both existing agents and authoring of new agents. Microsoft 365 Copilot licensed users continue to benefit from fair use of Copilot Chat or Standard harness agents included in their Microsoft 365 Copilot license. For all other usage, the Copilot Chat and Standard harnesses will continue to be billed using the existing fixed rate card. 

Get started today 

The GitHub Copilot harness is generally available now, and you can start creating these agents directly from the Copilot Studio homepage. We can’t wait to see what you build. Share your thoughts using the feedback control in the top right corner of the product. Your input directly shapes what we build—and how you build—next. 

Try Copilot Studio today!

Updated Nov 03, 2026
Version 5.0

12 Comments

  • The new Copilot Studio experience, powered by the GitHub Copilot harness, requires Copilot Credits for building, testing, and evaluating agents. In contrast, classic Copilot Studio requires Copilot Credits only after publishing. 

    Is there a way for MCTs or MVPs to get Free access to Copilot Credits so they can test agents in the new Copilot Studio experience? 

     

     

    • Paul Katz's avatar
      Paul Katz
      Brass Contributor

      You get 25,000 ($200 worth) if you purchase Partner Success Expanded Benefits.

  • czadras's avatar
    czadras
    Copper Contributor

    Hi Ryan, we have Power Automate flow calling Copilot Studio agents, using "Execute Agent and wait". Recently we are getting error "Sorry, something unexpected happened. We’re looking into it. Error code: InvalidContent." for agents created in new experience. Can PA flows still call new agents? 

  • LelandDriver's avatar
    LelandDriver
    Copper Contributor

    Hi Ryan, any way to restrict access to Github Copilot harness? Not all citizen developers are authorised to consume credits, and the standard harness is very hidden in new experience (under Other). Would love a method to route users to an appropriate harness.

  • Microsoft's AI pricing is becoming increasingly difficult to justify. It feels like you're paying multiple times for access to the same underlying models.

    • Buy a generic Copilot license.
    • Want Copilot in Power BI? You also need Fabric capacity.
    • Need an AI coding assistant? That's another subscription for GitHub Copilot.
    • Building AI applications? Azure AI Foundry runs on separate pay-as-you-go credits.
    • And now there are GitHub Harness credits as well.

    At some point you have to ask: why not just subscribe to Claude and eliminate all of this complexity?

    The fragmented licensing model makes it hard to understand what you're actually paying for, especially when much of it is built around access to the same family of foundation models.

    • Paul Katz's avatar
      Paul Katz
      Brass Contributor

      This is where the partner consulting community comes in, to guide customers on what they should implement. Because, yes, AI capability costs money to provide, no matter over which service it is delivered. It falls to us, the partners, to be the experts.

      I have a baseline Microsoft 365 E3 account on my tenant that I keep without Copilot User licensing, for demoing purposes. That's the starting point I tell clients they have. And it is pretty substantial; It automatically includes Enterprise-grade chat in M365 Copilot Chat, which can create and edit Office documents. It's also possible to create custom M365 Copilot Chat agents grounded in publicly available sites.

      At that point, it's about which purchase will provide value and which purchase won't. It might be that an organization just needs custom agents that get used a few hundred times a month. In that case it's all about Copilot Credits PAYG. For high use agents across a group of users, maybe they get the Copilot 365 User license. Is the agent going to be used externally a lot? Foundry Agents might be worth considering.

      Is the client doing something in Microsoft 365 super-duper complex? Well, at that point maybe GitHub Harness has a role, but that is a steep bar to get to.

      Regarding Copilot in Fabric Capacities, there is one feature that gets underrepresented. It is the only Copilot without Internet access and by design it will not use your business data in the model. Instead, it utilizes the semantic model to come up with the DAX necessary for the Power BI engine to run. From a security/compliance perspective, that alone might be worth the money. Or not. A Copilot Studio agent versed in a database schema can come up with SQL and utilize a SQL Azure database via a tool to do some pretty impressive stuff.

      And yes, Github Copilot costs can now easily add up.

      On the plus side, this raises the value of our consulting work. Organizations need to be much more disciplined about how they deploy AI compared to other recent technologies. They need experts, namely us.

  • Paul Katz's avatar
    Paul Katz
    Brass Contributor

    Ryan, I applaud your team on the hard work you all have done. I'm sure through the preview your team has learned as much as we have, perhaps a big one being the sheer cost of powering these incredible models and the outcomes they can create.

    That said, I write here to ask for a reconsideration of the branding structure. Copilot is a brand embedded in the notion of a helpful sidebar or chat agent. Copilot Studio is a maker low-code brand, rooted in the 365 ecosystem. This new "GitHub Harness" branding, a developer brand, muddies the waters. I can't make it intellectually or emotionally fit with Copilot Studio. Your own blog piece above, however, offers a solution.

    You say this new GitHub Harness technology is what powers Copilot Cowork? Fantastic. Guess what you're creating with Copilot Studio when you choose it? A Copilot Cowork Agent. The natural next autonomous step from Copilot Agents, the type of agent you now call Standard Harness Agents.

    Forgive me while I use these new brands as I continue to describe them; if nothing else, I hope you will see how natural they sound. Copilot Cowork Agents and Copilot Agents overlap somewhat in capability, but truly support distinctly different production scenarios.

     

    Copilot Agents provide helpful smart chat agents, what you use to quickly guide customers as a virtual customer service rep. Or provide quick answers to folks on a Teams call. Or through M365 Copilot Chat, allow for completion of impressive yet somewhat light tasks. Due to the "help in the moment" nature of Copilot Agents, they can be deployed on websites, in Teams, in M365 Copilot Chat, and also in various other social and communication platforms. Doubling down on their helpful nature, they can also be offered to customers on other Microsoft 365 tenants, projecting helpful support beyond the boundaries of one's own tenant and into another.

    Copilot Cowork Agents are agents that take a general mandate, reason multiple steps necessary for completion, and then execute and complete what is asked for, even if it means writing and executing a whole program! While they often have a chat front-end for helping users think through what to eventually ask for, their strength isn't in what it does in a chat but what it does while their master(s) get something else accomplished. Due to Copilot Cowork Agents' impressive computational needs, unlike Copilot Agents, they are billed as a metered capacity and generally aren't designed to be externally facing. However, compared to the cost of manually doing everything Copilot Cowork Agents do, the price for running them can be trivial compared to the time and money they save.

     

    There you have it. Two solid brands that don't cannibalize each other and succeed in conveying that the new product serving a different need won't be tossing out the old product that has a valued different place in our toolchest.

    In the name of all the good work your team has done, I request that Copilot Agents replace "Standard Harness Agents" and Copilot Cowork Agents replace "GitHub Harness Agents". Your outstanding work deserves outstanding branding ;)

    I am at your disposal if you wish to discuss.

    Thanks,

    Paul

     

     

    • pmb90's avatar
      pmb90
      Tin Contributor

      Great work! But I agree with Paul, change the branding/naming. It will create even more confusion for our clients.